maxgadney.com

Sentiment Analysis hocum

What could he mean?

In last weeks Economist, there's a semi-decent article on reading sentiment from large masses of anecdotal data (tweets, mainly).

It's only semi-decent because it doesn't go the whole hog and say 'this doesn't work yet'.

Many smart senior people (un-named because their companies are involved in this snake-oil sale)  have confirmed this for me, saying "sentiment analysis software? don't bother"

The number of pitches from software firms I have attended in the last few years promising automated sensemaking of data says more about my patience than their products.

Not all is lost...

The data that twitter can supply is good for basic mentions, the basic smell of the situation - just not it's meaning -good/ bad, action required, not required. It's a canary in the mine, but it can't tell you what gas it is. This what many brand/ marcomms research dashboards supply - but many aren't optimised for the 'now take action' moment - they are mostly like webcams of rivers - an ongoing situation goes on.

What does work is smart , linguistics trained researchers, alerted by flagged moments in data inputs, then raking through data/ tweets to find meaning. Machines are useful for sorting these out but real meaning comes from human analysis still. Your engineers could help by writing some scripts that help the sifting/ alerting.

Everywhere data is used for analysis, a human is required for the real interpretation. The robots just aren't that smart yet.

Job done, but which one?

JWT have teamed up with Millward Brown research to visualise the annual MB 'Brandz' study.

They've done it by visualizing the brands as characters. Some work some don't. But it is cute. It's a bit like the schools in schooloscope (but their limiting of characteristics made them more legible)

As visualization it us questionable but I don't think that is the purpose.

Millward Brown are known for being a bit dry (though very reliable) and this sends a message that they are addressing that, or at least reaching out to ad planners who like this sort of thing, a key audience for them.

So it seems to fail in it's primary purpose of info dissemination but I don't think that was the prime purpose.

It's like the google Im feeling lucky button. It is a SUPFTAFABJ.

A seemingly useless product feature that actually fulfills a branding job.

By having it there, Google are saying, "We aren't that serious - we're Gooogly!"

And that is what MB are trying to do I think. 

Just because it is the pictures...

News from the Daily Mail, UK (3m 'readers')

I have just put my name to this piece on the Nieman Foundation for Journalism.

It basically says 'stop doing shit infographics'.

You can read the more reasoned argument for yourself.

I would add to their words the following:

Thinking that you can elaborate rumours with a graphic is insane. Graphics or illustrations make clear what was previously not clear. Keep any vagary textual.

The English press has spent a good century using our highly flexible and nuanced language for some of the most scurrilous and underhand speculation which stopped short of actual slander only because of deft opacity of the writing.

You can't do that with pictures. They expose you and commit you to showing. Which is possibly why the english press are so bad at using pictures. But that is another story.

Well done Juan and Alberto.

Visualisation, HypeCurves and Years of Dispute to come

Having highlighted Stephen Few's blog yesterday, an ex-BBC colleague, the super smart Lulu Pinney, pointed me to this discussion, where Few is manning the barricades,Rourkes Drift-like, against thousands of oncoming David McCandless fans as they rush screaming towards him.

He seems to deal with them pretty well (and for any 2000AD/Slaine fans it seems "he didn't think it too many")

Lots of metaphors there.

But the main thing is that Visualisation is firmly climbing the hypecurve. It takes public figures like David, (who I like and respect) to bring a new popular way to the table (not just Doing the work but Publicising it too - which he does very well). Without this approach and work, there would be no hype. Hype needs  a product, a face and point of view, an idea that is spreadable if tangible or not.

What happens next is that professionals like Few (or Nielsen/ Norman, before) (who are often more versed in reasoned debate rather than design-blog/ you tube emotionalism ) then point out that this stuff could benefit from some other (user) perspective. Their tone often rallies their opposite numbers who see thhem as aloof and outmoded.

Then there is a scrap that goes on for years. And it is a fight invisible to the general public.

The design press/blogs then move on to something else and declare "x is rubbish, as the plane loaded with all the counter arguments crashes groundwards.

Then nothing. Well, not nothing, just a period of no hype, heat debate, just lots of doing and practice, as people influenced by all camps - and others too set about their careers and making.

A few yesrs later, the industry has assimilated all the arguments at a wiser level and can draw on ideas of utility as well as engagement.

Absorbing new and counter-intuitive ideas takes time and it is natural to argue/ disagree at first, but just you wait. Soon we'll have fantastic data apps as common place just as we have largely good and usable websites as standard where we didn't ten years ago.

And it is all down to these different voices entering the debate, their messages gradually being absorbed and digested  into something good. Long may it continue. And it will.



A Few good men

This looks to be a very good blog from Stephen Few.

In the same way a generation of 'web designers' were infused/mentored by HCI veterans. (respect as always to Jakob Nielsen et al), we may be seeing a similar thing for the data vis community.

The recent generation of young turks is doubtless having fun with data scrawling but at some point it will pass people by unless there is a purpose or utility to it. They've got the engagement sorted. These things are mostly usable. What they are not is useful.

That is where peopel like Few come in. They work in analytics - using data for decision making. They are ideal real-life mentors, solving real life problems. They can point the way to thinking of these apps as tools for whatever outcomes.

So, read all of his blog - all the archives too (i will) before you go anywhere near popcorn-data-blogs again.

Solution in search of a problem

Just thinking as to why I shudder when asked to Interact with many visualisations - unless it is a pet interest or I know what I need from it, being asked to 'interact' with something throws me back to the days of 'hidden navigation' and other such bollocks.

It's because these things re put together by experts and they expect you lot to have the same level of interest in their often way too clever take on stuff.

"Lets the user explore" - well, you could do , like you could shw someone all the parts of the toaster rather than selling ot to them

I'm a bit more into task based stuff now anyway - tools that help people with business problems - people that need to show data as proof/evidence (in market research since you ask) . Maybe other people need such tangible problems too.

A colleague said of one visualisation - "its a solution in search of a problem".  

The march of time - onwards and upwards

Very decent timeline from the guardian here.

- makes real effort at resolving the far out/ deep in of time lines - id like to see it stretch each way in and each way out even more - it is a little bit on rails - id like to see the whole

- v clear iconography (although the clear colours would really illuminate the Whole from a far out view - and id like toi see more links across countries (maybe there aren't any?)

- nice metaphor of time as a zoom forward - shame the mouse controls cannot use scroll wheel - you 'go' forward but still need to move left to right at the top

overall really good - and a positive step forward

The Story of Economics - Blastland's new work

Michael Blastland was one of the most popular speakers at the recent Design of Understanding conference.

Here is his new radio show - starts today. Iplayer available to UK people after.

 

Make graphics simple, not simplistic

 

Hans Rosling said to me "graphics won't do everything - they'll simplify some points, they'll introduce the idea and help to highlight important areas, but for some of the more complex explanation like causation or correlation, they have their limits".

That is borne out in the comments of this piece on redesigned health reports in Wired. 

I personally think the graphic pieces are good enough to illustrate the point, but probably aren't to a productised standard. You could say 'yeah but it was a magazine deadline' but you could say 'don't give lame excuses'. Both sides are right. (and i still buy Wired).

No one is going to disagree with the idea that health data could be given visualsiation attention and could be made more useful.

But the reality of doing anything good when dealing with this stuff is that it takes a tonne of work to understand the problem first.

In research at the BBC, a repeat issue was the fine line between making things simple (good, useful,accessible, focussed) and simplisitic (glib, generic, useless).

 

Are comics a bit crap?

I love comics. At least I used to before Monday.

I'd finished reading The Passage - a well written, non-Twilight, vampire horror novel. It was great. The story world was compelling and there were some good characters and a ripping story. A real page turner.

When I finished it, I thought how good it was and also how long it took - 750 pages over 2 weeks. It was densely packed and involved.

I thought of my favourite comics, say The Punisher, In The Beginning, by Garth Ennis. I think on balance, Punisher fans would have got more out of a 750 page written novel. Think of the amount of character exploring you could do. Nicky Cavela's backstory, the Company's deliberations etc. Could have been fantastic, but it was a comic and ideas in comics can only go so deep.

What we get in a comic is impact - and accessible engagement. It's not just the digestibility but the jaw dropping punch that you get (especially in Ennis' pacing) when you turn a page. Books cannot do that, but on balance I would rather have the slower burning effects of a page of prose and their lasting effects than an impactful end panel that lands it's case quickly but forgetably.

For those of you that know me and my love of comics will know that I am a bit disturbed by all this - but to continue -

I'm thinking the same with information graphics - are they just adverts for the text experience?(When there is one) If they are useful software then they are obviously an experience in themselves.

Pictures have momentary attraction but a thin capacity for information and importantly, ideas.

Prose on the other hand absorbs ideas like blotting paper and retells them with a lasting sophistication.